##################################################################
# This is benchmark code for using titanic
##################################################################

#This will load data and help create a output function
#attaching titanic data
attach("/titanic/input/titanic.RData")

head(train)
head(test)

library(h2o)

#intialize h2o on your server
h2o.init(ip="", port="")

#loading data into h2o
train_hex = as.h2o(train)
test_hex = as.h2o(test)

nrow(test)
nrow(train)

#Feature Engineering
# Analyze your data and take care of missing values

#Random Forest
titanic.rf = h2o.randomForest(y = 2, x = c(1,3:15), training_frame = train_hex)

#Look at the model
print(titanic.rf)

#Prediction using h2o
predictions <- h2o.predict(titanic.rf, test_hex)

#Convert h2o frame to R frame
prediction = as.data.frame(predictions$predict)
head(prediction)
summary(prediction)

#Normalized Data
normalized = (prediction-min(prediction))/(max(prediction)-min(prediction))

#Histogram of example data and normalized data
par(mfrow=c(1,2))
hist(prediction,xlab="Data",col="lightblue",main="")
hist(normalized,xlab="Normalized Data",col="lightgreen",main="")


#Creating a submission frame
test <- as.data.frame(test_hex)
passengerId <- test$passengerId

#Creating a submission data frame
submission <- cbind(passengerId,normalized)

#Write submission
write.csv(submission, file="taposh_test_submission3.csv", row.names = FALSE)